I purchased NCEP reanalysis GRIB data for summer 2001, downloaded the PC HYSPLIT model, and figured out how to convert the GRIB format to the HYSPLIT format using the compiled fortran program: ncr2arl. However, I am unsure about what latitude and longitude (lat/lon) to input. I read that lat/lon are at the center of a 100×100 grid that is extracted from the GRIB file. So if my study area is around the North Eastern Pacific/West Coast of the US, should I just input my center lat/lon from my study area? I am trying to do backward air parcel trajectories of the marine atmospheric air located off Oregon. I am worried that if the grid extracted is only 100×100 from my center lat/lon that long duration backward trajectories might cover a larger area than the grid extracted. How should I deal with this? Finally, when using display.exe (or HYSPLIT GUI interface) to display contour maps, should I again enter the center lat/lon of my study area (in other words is the diplay.exe plotting the extracted 100×100 grid?) and what is meant by the radius (degLat)? What value of radius should I put in and how does different values change my contour map?
Yes, the lat/lon are center points. Output grid spacing is 100 km. As you describe, the trajectory could go off the grid depending on the duration of the trajectory, the winds and the grid centerpoint. One way around that is to make more than one grid. Make sure they overlap. Then you can specify all the grids in HYSPLIT and when the trajectory reaches the edge of one, it will continue on the next one.
For display.exe, you enter the center lat/lon and radius of the area of the meteorological data you want displayed. Yes, the plot is in a square shape, but imagine a circle overlay – use that radius. If you had a global dataset you can use ‘display’ to show the relatively small portion of the grid you are interested in instead of the whole grid. Maybe you want to see the whole grid. In that case, you would enter the center of the grid and the radius in degrees latitude. Watch out if the radius is too large so that the size of the plot is not larger than your grid domain because the program may crash. Try out a few different radius values to get what you want.